1. Field of the Invention
The present invention is related to search engine technologies and more specifically to human-assisted search engines. A method and system for improving the relevance of search results is disclosed.
2. Description of the Related Art
In general search engines are keyword driven systems. Keywords are generated from a user request and matched to target documents, advertisements, etc. This practice is generally established by such services as Google® or Dogpile®. The use of keywords allows a search engine to produce relevant documents by using various methods of ranking documents, etc. An algorithm is applied to the content of each document and a ranking is assigned. When a query is entered by a user, the search system analyzes the query to extract keywords contained in the query and presents documents in an order related to the ranking.
In the case of an automated search, the algorithm used to determine relevance may be generalized in order to search various types of documents. The algorithm used is designed to approximate human judgment. Such systems have enjoyed commercial success.
However these systems have weaknesses. For example, a more specific query may not produce a more specific result. The number of keywords in a query will typically increase with the length of the query, and the relative strength of any keyword may decrease. Relevance of results may be affected by factors such as search engine optimization. If an algorithm for determining relevance is known, a website designer may optimize his or her website to improve the website ranking. Relevance is not a static property of a resource. Relevance may be affected by time and/or other factors which can not be deduced from keywords. A typical indexing strategy may be limited by the dynamic nature of sites, such as game playing sites, auction sites etc. Such resources may be easily accessible to a person, but are not easily indexed by a search engine.
In the case of a human-assisted search engine, human judgment may be used to determine the relevance of a result or resource. This technique is used by open directory projects such as DMOZ. Such systems are mainly constrained by the need to have a large number of human editors. For example, DMOZ had 75,000 editors, 500,000 categories, and 4,800,000 websites in its index.
In light of the above and other existing problems, a scalable method and system for producing relevant targeted search results responsive to a user request which is based on use of human judgment and intelligence would be greatly appreciated.